FRM L2 Feedback

Discussion in 'About FRM' started by Jiew Kwang, Nov 19, 2011.

  1. mcarthur724

    mcarthur724 Member

    41) Gross incomes for previous 3 years , outsoursing commisions for previous 3 years, and profit from equity sale in last year are given. What is the ORC(BIA). The problem is there is no answer if we use only gross income data. So we should use another data somehow. Pls explain.

    For this question I just took average of Gross Income for the 3 years and ignored all of the other additional expenses that they gave because BIA just uses Gross Income. However, the part I wasn't sure of was how to deal with the negative Gross Income in year one. First I tried to exclude the negative year 1 GI and just take the average of the positive years 2 & 3, however there was no answer choice for this. So then I included the negative year 1 in the 3 year average for GI, multiplied that by 15%, and there was an answer choice for this so I went with it. Does anyone know if this is the appropriate handling of a negative GI in a year?
  2. Quest4FRM

    Quest4FRM Member

    In BIA approach, one ignores the negative income from numerator and hence denominator will be average of the remaining incomes i.e the sum of the 2 positive gross incomes multiplied by 15% and divided by 2 (not 3)
  3. Quest4FRM

    Quest4FRM Member

    For LVAR, the formula is LVAR (exogenous) * LVAR (endogeneous)...but i think th question mentioned that the trade did not affect the market value or was small compared to the market...something in that i used only LVAR (exogeneous)..i might be wrong given that i didnt read the question thoroughly...
  4. AG

    AG Member

    @McArthur, yes you are right about the cumulative PD problem... I tried that approach, however I do not remember getting anything close to the options. So I chose Cumpd(3)-cumpd(2). The options were 10.sumthing, 11.sumthing (which was cumpd3-cumpd2), 16.sumthin and don't rem the other option.

    To get the ES you just need to average out 96th-99th percentiles. there was an option which matched. I tried to incorporate the 0.5 by assigning 1/2 weight to the average of 95th and 96th but didnt get anything close to the options.

    Regarding the questions on regression based data, one was straightforward, since the beta was > 1, the fund underperforms in downward markets.
    The other one was a little bit tricky since they added -0.4*max(0, benchmark return)... so in an upward market the beta would be leass than 1 (1.2-0.4) and in wa downward mkt the beta will be >1. So the fund will always under perform. I also ignored the regression estimates. In one case the alpha was very low and in the other case alpha was not statistically significant.


    42) answer was 20: 400,000 divided by max daily of 20,000 contracts
    43) Regarding the exo endo formula, I dun rem. But I think in one of David's practice problem he did LVAR(exo)*(LVAR(endo)/VAR ratio) and this makes sense also, since if you just take LVAR(exo)*LVAR(endo), you have double counted the VAR.

    @Quest4FRM, you are right about the impact of the trade was low (but not exactly 0). Otherwise the elasticity would have been zero. However, they put a non-zero value for elasticity. So i kinda included that.

  5. Jiew Kwang

    Jiew Kwang Member

    I did what Quest4FRM did for LVaR cuz I did see that firm owns 1.5M shares out of 500M shares. So I just used exogenous spread. Also agree with the BIA approach using 2 to divide the positive gross incomes. The other information given were red herrings. So many of these type of questions. It's so much 'funner' than level 1 this May..
    AG, my answers were also quite similar to yours only differed in a few questions.
  6. Quest4FRM

    Quest4FRM Member

    Well, yea...for that LVaR question i can see the logic both ways...i got an exact answer using just the Exogenous approach...but knowing what GARP does, it could very well be a trap ;)... Maybe if AG got the exact answer using his approach and it matched the answer options then he cud very well be right...

    For that default intensity one...i too used (Cum (3) - Cum (2))/(1 - Cum (2)) and i think i got answer as one of the answer options...

    There were a few questions on extreme value theory as well-

    What happens when we increase the threshold...think one of the answer options were it converges to a common distribution (GPD)
  7. AG

    AG Member

    Quest4FRM, I think LVAR_exo was the right answer since by my approach I did not get the exact option of 566,000, this number was the closest value to my answer. So I marked it. I guess you just had to mark LVAR_exo. Clearly a time trap I ignored ;)

    I agree with ur response to the GPD question.

    Guys do you rem the question on VAR back-testing at 95% and 99% - I chose: the inference drawn at the 95% back testing will be more reliable than the 99% test.

    One question on sovereign CDS spread: I marked the estimated expected loss will be a large part of the spread. But I guess I was wrong.

    One question on two assets portfolio. Asset A - 2 mil, asset B - 1 mil. VAR of each position is 40,000. rho=0.5. What is the increase in VAR if Asset A has 1 mil and B has 2 mil. Think level was 95%. Forgot the answer.

  8. Jiew Kwang

    Jiew Kwang Member

    I can't remember the options clearly for the back testing question. I skipped it in my first run as I wanted to do the straightforward questions first at the start.

    Sovereign CDS spread i picked the option that said "spread almost exactly..." (was torn between this option and AG's)

    VaR qn:
    sqrt(40^2+40^2+2*0.5*40*40)=Diversified VaR of initial portfolio
    Next find the new portfolio VaR with the new weights
    Find New VaR = SD (New) * 1.645 = New Diversified VaR of portfolio
    Take difference. I got the exact answer in 1 of the options but forgot the number...
  9. AG

    AG Member

    Hey JK, do u rem the question: A firm needs a loan funding. If its credit rating worsens so that it is difficult to get a loan, which of the following transactions it could engage in... Options CDS, CLN, etc... I marked CLN
  10. Quest4FRM

    Quest4FRM Member

    Yea for the LVaR question, as far as we all got the same answer of 566,000 it really doesnt matter i guess :)

    For the 95%, 99% backtesting at 90% confidence, i think options 3 and 4 were of those mentioned type 1 error and the second option also referred to the same thing without explicitly mentioning type i thot those 2 cudnt be the answers...out of the remaining 2 options...the first option mentioned that we (definitely) accept the 95 model bcoz it is more reliable...i guess the 95 model no doubt provides us more data points to validate the model compared to 99...but reducing the confidence level will also make the model less i went with the 2nd option which said accept or reject the 95 model bcoz... Dnt knw whether my reasoning has a well defined logic or not, but i marked b as the answer option...

    For the sovereign spread question, i read in one of the readings that spread is largely due to unexpected loss (risk premia that investors associate with socereign bonds) and that expected loss is only a small proportion of the sovereign spread...hence i selected that expected loss is a very little part of spread...dnt knw if i am right or wrong bcoz there was data abt LGD, PD etc which was not at all used by me...

    I didnt have time to calculate that 2 asset portfolio var question...randomly guessed something...

    There was another question on country defaults...which of the 2 statements is correct types...40 out of 200 countries have never been under default or restructuring and one more statement (cant remember)

    There was another on VaR...removing which asset will reduce portfolio VaR by max...i think there were 4 assets and a table given...with 4 or 5 columns...didnt know which column to use to make the decision :(
  11. AG

    AG Member


    You are correct about the sovereign spread question. I cross checked after the exam.

    On the country default I think I marked this one only... 20% have not defaulted.. The other option was wrong I guess.

    On the VAR side, I guess the marginal VAR colum was important, not sure though. I chose the one where the marginal VAR was 19.sumthing.
  12. AG

    AG Member

    Also, does anybody rem the question on convertible arbitrage... Source of return or sumthing when it is delta neutral.
  13. Quest4FRM

    Quest4FRM Member

    Yea for the country default think i also remember having narked this statement as true...the other one was wrong...but cant recall what the statement was..

    Oh yea..there was this question on convertible arbitrage...dnt remember the was something related to volatility increase...
  14. Jiew Kwang

    Jiew Kwang Member

    Hi AG, if i did not remember correctly, I marked CDS for that option. Because CDS is the only unfunded instrument. I can't exactly recall the question now though so I'm not sure if I am even on the same page as you. I can only vaguely remember those convertible delta neutral questions. There were just too many of them.. LOL.
  15. Quest4FRM

    Quest4FRM Member

    Haha..i think we all marked different answers for that question...i marked TRS...i also dnt remember the exact question...think it was that the firm needs to buy an asset but its standalone creditworthiness is not good enough for it to get a what can it do to get that asset...with TRS, the firm can approach the bank to buy the asset on its behalf and pass on the return of the asset to return the firm will pay some interest rate (Libor plus something)...which is equivalent to owning the asset at a price of libor plus something...
    But i am not sure what the exact question was
  16. EIA

    EIA Member

    On the QQ plot question, this is what Kevin Dowd said "

    Third, if the empirical distribution has heavier tails than the reference distribution, the QQ

    plot will have steeper slopes at its tails, even if the central mass of the empirical observations

    are approximately linear. Figure A3.3 is a good case in point. A QQ plot where the tails

    have slopes different than the central mass is therefore suggestive of the empirical distribution

    having heavier, or thinner, tails than the reference distribution.
    The diagram on the question was the same diagram on the text.
    From the above, Thin Tail is also correct.
    David what is your take?
  17. AG

    AG Member

    Amazing to know how you guys came up with different logic and al of them seem to be correct.
    My logic - The firm needed sumthing equivalent to the loan i.e. it needed funds. CLN was the only option through which the firm could get funds since it's equivalent to a funded CDS. Also investors are short a CDS in CLN, so they want to bear the credit risk. CRAZY.
  18. Quest4FRM

    Quest4FRM Member

    @Emeka: r u saying that QQ plot was not referring to fat tails...?? Sorry but i cudnt quite understand the reference that u have posted without the diagram..and i dnt exactly recall the diagrsm at this point..
  19. Imad

    Imad Member

    I did the same as you
  20. Quest4FRM

    Quest4FRM Member

    @AG: haha...yea..thats the fun part of FRM..sometimes all options look correct and sometimes none looks correct :D

    The thing with that question was that the answer depends on what exactly the question was referring to...did it talk about what the firm cud do to get the same return from the asset in spite of reduced creditworthiness or what cud the firm do to own/purchase the asset...i think the answer shud be TRS in the former case and CLN for the latter, since in a TRS the firm does not own the asset...

    Unfortunately i have forgotten the exact wordings of the question...

Share This Page